Borders of Compassion: Immigration Preferences and Parochial Altruism

Resources

Figures & Tables

Authors: Alexander Kustov

Journal: Comparative Political Studies (2021)

DOI: 10.1177/0010414020938087

Abstract

Anti-immigration preferences among educated and racially egalitarian voters is hard to explain using existing frameworks of self-interest or prejudice. I address this puzzle by developing a theory of parochial altruism, which stipulates that voters are motivated to help others at a cost, but they prioritize helping compatriots. I hypothesize that parochial altruists or voters high in both “nationalism” and “altruism” are more supportive of immigration restrictions perceived to be in the national interest. However, parochial altruists are also expected to be more supportive of increasing immigration when it benefits their compatriots. I test my theory by conducting a population-based UK survey. Using a novel measure of elicited preferences, I first find most altruists who donate to domestic rather than global charities are as anti-immigration as egoists who do not donate at all. Using a conjoint experiment, I then show voters support increasing immigration when these alternative policies benefit their compatriots.

Keywords

immigration, redistribution, public opinion, survey design, experimental research, altruism, nationalism

1Jackson Institute for Global Affairs, Yale University, New Haven, CT, USA

Corresponding Author: Alexander Kustov, Postdoctoral Associate, Jackson Institute for Global Affairs, Yale University, 55 Hillhouse Ave, New Haven, CT 06520, USA. Email: alexander.kustov@yale.edu

Introduction

When asked about the most important issue facing their governments, a plurality of EU and US voters in 2018 chose “immigration.”1 The political conflict over human mobility, however, is not new. Indeed, humanity has grappled with the issue of migration since the very inception of state institutions—both citizens and politicians alike have often been divided over who gets to be a member in their community. A large literature on the topic notwithstanding, it is still unclear what exact combination of factors motivate this conflict and whether there can be any democratic policy compromises preferable to the status quo. The modern-day highly restrictive government regulations on immigration2 are economically inefficient and harmful to the global poor.3 However, most voters in rich countries oppose immigration.4 While this opposition is well-documented, the sheer extent to which most voters dislike any increase in immigration is often overlooked. According to my representative UK survey used in this paper, only 20% of educated and (racially) liberal voters— supposedly the most pro-immigration group—preferred increasing immigration (Figure 1). Such widespread opposition to increasing immigration among even college-educated and racially egalitarian voters is hard to explain using the major existing theories that attribute these sentiments to ignorance and

Figure 1. Immigration preferences across various groups in England. Note. Based on the England Qualtrics sample (2018). For details, see the Data section and Online Appendix.

prejudice. This is especially puzzling given that political behavior is often motivated by prosocial concerns rather than just self-interest.5 Why do voters—including those who genuinely care about others—support anti-immigration policies that severely restrict individual freedoms and leave the world worse off? More practically, under what conditions, if any, would most voters accept fewer immigration restrictions? This research addresses these questions by reconsidering public attitudes toward immigration as a function of parochial altruism, a stable individual predisposition that motivates political behavior. In particular, I develop and empirically test an original decision-making model stipulating that voters often face an altruist’s dilemma: they are willing to help others even at a personal cost (“altruism”), but they prioritize helping their fellow citizens first (“nationalism” or, to be more precise, “national bias”). As a result, independent of their self-interest and other concerns, parochial altruists are expected to oppose or support increasing immigration or any other policy if they believe it hurts or helps their compatriots. Given widespread public perceptions about immigration’s negative national impacts, the theory implies that voters, and especially those who are altruistic, tend to favor harsh immigration restrictions that they perceive as necessary to secure the well-being of their compatriots. I thus hypothesize that respondents high in both altruism and national bias (“parochial altruists”) are the most opposed to immigration and respondents high in altruism but low in national bias (“universal altruists”) are the most supportive.6 However, the theory also implies that parochial altruists could still conditionally support increasing immigration when these alternative policies explicitly benefit their compatriots. At the same time, I expect more altruistic voters— regardless of their national bias—to be more supportive of greater domestic redistribution, which generally poses no conflict between national and global interests. Unlike related “social identity” and “sociotropic” accounts also predicting that voters choose policies they think are better for their countries (Hainmueller & Hopkins, 2014), the proposed framework goes further and specifies when this would or would not be the case as testable hypotheses. In doing so, it clarifies that national identity and sociotropic perceptions largely matter due to selective altruistic incentives rather than merely self-interest or outgroup animus. Furthermore, it offers a novel prediction of a systematic interaction effect between individual group bias and altruism in explaining political attitudes and behavior. To supply initial evidence for my argument, I use existing cross-national survey data and document the surprising non-relationship between self-reported altruism and immigration attitudes in most countries. I then show

that altruistic voters can in fact be either more pro-immigration or anti-immigration depending on how much they favor their compatriots over foreigners. In line with my argument, this implies that “nationalist” voters are much more opposed to immigration when they are altruistic. To address social desirability and endogeneity concerns, I then build on this evidence and conduct an original representative study with incentivized economic games and choice experiments in the United Kingdom. First, using a novel incentivized measure of elicited altruistic preferences, I find that almost a third of voters are willing to contribute to domestic charities at a personal cost and at the expense of comparable global charities. I then show that altruists who choose to donate domestically rather than globally (“parochial altruists”) are as anti-immigration as egoists who do not donate at all. But parochial altruists are more supportive of policies increasing immigration than egoists when these policies explicitly aim to benefit their compatriots. Second, I complement my survey with a pre-registered conjoint experiment to test the counterfactual implications of parochial altruism against existing theories. In particular, I consider the two prominent accounts linking voters’ motivations and policy preferences—pure self-interest and taste-based ethnoracial prejudice. To do this, I estimate the effect of perceived policy consequences for personal and various collective interests, as well as the number of immigrants and their origin, on immigration policy choice. As opposed to existing conjoint and other survey experiments on attitudes toward various individual immigrants, this design can examine the relative importance of various (potentially correlated) motivations that underlie preferences over immigration policy in general. According to the conjoint results, most UK voters could support increasing immigration from non-European countries if they believe it is beneficial for them and their compatriots. As further indicated by my similar findings in the US, the political dynamic of parochial altruism generalizes across contexts. These results have important implications for immigration public opinion and policymaking. Most prominently, can governments realize the potential gains of immigration while still being responsive to (evidently anti-immigration) citizen preferences? If these restrictive preferences are largely conditional on the perceived national effects of today’s policies rather than the categorical rejection of foreigners as indicated by the evidence here, a political compromise on immigration should be feasible. Given the apparent failure of efforts to change negative perceptions of immigration, however, my results suggest that a more effective government strategy is to consider alternative, less restrictive policies that more explicitly and straightforwardly benefit average citizens alongside with potential migrants.

Theory and Hypotheses

Why do voters support or oppose immigration? To answer this question, we need to consider people’s basic motivations in politics. Building on recent advances in economics and psychology (e.g., see Fehr & Schmidt, 2006), numerous behavioral and attitudinal outcomes in political science are increasingly seen as a function of altruistic motivations beyond material self-interest and other considerations. Most prominently, the genuine concern for the well-being of others has been shown to motivate political participation (Dawes et al., 2011), support for redistribution among the rich (Gilens & Thal, 2018), and preference for global climate cooperation (Bechtel et al., 2019). Similarly, some scholars argue that, if it exists at all, altruism is exactly what motivates people to support immigration, especially given the plight of many less fortunate immigrants. After all, many people do view policies increasing immigration as an act of charity (whether rightfully or not). In line with this argument, there is some evidence from self-reported surveys that altruistic people are generally more accepting of international human mobility (Dinesen et al., 2016; Hansen & Legge, 2016; Newman et al., 2013).

A Decision Model of Parochial Altruism in Political Behavior

Despite its insights, however, the literature described above has largely neglected to address the fact that altruistic motivations are rarely impartial. In other words, altruism is often “parochial” (Bernhard et al., 2006)—that is, people are more willing to sacrifice for some more than others based on their group membership. Accordingly, there can be at least three distinct ultimate motives for cooperation or conflict, the last two of which are often confused in the literature: (egoistic) self-regarding, (parochial) group-regarding, and (universal or impartial) human-regarding, which respectively aim to advance perceived self-interest, group interest, and global interest (Schwartz-Shea & Simmons, 1991). While these motivations and their distinctiveness have been widely examined in a lab setting using multilevel social dilemma games (Dreu et al., 2014), they are likely even more relevant in real-world politics. To capture this dynamic, I provide a succinct decision-making model of parochial altruism in which more or less parochially altruistic individuals choose between various policies depending on their personal and various collective consequences.7 More formally, let k be some policy or political participation alternative that differentially affects total (T), group (G), and individual (I)8 payoffs relative to the status quo.9 The utility function Uijk of agent i in group j for k can then be presented as follows:

Figure 2. Graphic representation of possible utility weights in the parochial altruism model.

U I G T e I G T e ijk ik jk k i i ik i ik i jk i i k ik (,,,,,) (), Î± Î² Î± Î² Î± Î² = + + − − + 1

where 0 1 1 < < + < Î± Î² Î± Î² i i i i,,. Here Î±i and Î²i are the relative weights of agent i for valuing the (relative change in) well-being of herself and her group over the global welfare in general (also see Figure 2). The residual term eik stands for other relevant factors that may be specific to each individual or policy alternative. Finally, I assume that the level of concern about relative well-being is generally decreasing from self-interest to group interest and global interest: Î± Î² Î± Î² i i i i > > − − () 1. In other words, most people are expected to prefer a positive change in their own payoff to a similar change in their group payoff and only then the global payoff even though the latter is always much larger in absolute terms.10

The model thus acknowledges that, self-interest (Î±i ik I) and other concerns (eik) aside, people have mixed prosocial motivations that may be in conflict with each other. Consequently, altruistic actors with sufficiently greater ingroup weights may “rationally” decide to harm others when there is a tradeoff between group and global interests, even in the absence of intrinsic negative weights (e.g,”outgroup hate”)11 in their preferences. While the

literature has traditionally examined a potential conflict between selfish and collective interests, many problems in politics involve a choice of helping some groups at the expense of general social welfare and self-interest (or sometimes even harming other groups). Consequently, it may be hard to sustain cooperation and avoid conflict even in a world of prosocial individuals. Although the model is agnostic regarding the relative importance of various group interests for individuals (Î²i j G), in advanced democracies today parochial altruism must be more directed toward compatriots than any other non-familial ingroup members. After all, national citizenship is associated with more institutional and coercive authority relative to other social categories (Schildkraut, 2014). It is also one of the most salient, albeit often taken for granted (Billig, 1995), social self-concepts that guides and determines the whole life of an individual (Goode & Stroup, 2015). Relatedly, nationality or citizenship can be viewed as the most consequential ascriptive social category for one’s material well-being (Kochenov, 2019) with 66% to 73% of global variation in individual household income (PPP) determined by one’s country of birth. In this sense, nationality is more influential than all other ascriptive characteristics such as race, ethnicity, and gender (Milanovic, 2015). Finally, citizenship—as opposed to other ascribed identities—is one of the only circumstances of birth for which discrimination is, albeit variably, legal and socially encouraged (Brubaker, 2015). At the same time, strong norms against ethnic parochialism in developed countries may significantly constrain the expression of these motivations (Tankard & Paluck, 2016).

Implications of Parochial Altruism for Immigration Attitudes

Despite the common assumption that altruism always makes people supportive of immigration, the current contentious debate over the issue provides a quintessential example of how ambiguous the perceptions of “social good” can be in (inter)national politics. On the one hand, harsh immigration restrictions are inefficient and they harm potential immigrants (e.g., Pritchett, 2006). On the other hand, people may feel especially compassionate toward fellow citizens—who are often perceived to be harmed by immigration—and resist international mobility despite their personal interests or the global benefits involved. While this logic is more apparent regarding the distributive consequences of economic activity, it should similarly apply to non-material factors of well-being related to differences in norms and values. The available evidence notwithstanding (for a review, see Blau & Mackie, 2016), the effects of immigration on receiving countries and their citizens are predominantly portrayed as negative in the media and correspondingly perceived

as negative among voters (Muste, 2013). Although voters are probably more optimistic about the effects of immigration on (potential) immigrants themselves and the world as a whole, these beliefs should only matter for those who care sufficiently about the well-being of foreigners. As a result, conditional on widespread beliefs about negative national impacts of immigration in advanced democracies, I derive the following observational hypothesis:

Hypothesis (1a: Altruist’s dilemma) Compared to egoists, people of higher parochial (universal) altruism are less (more) likely to prefer policies increasing immigration.12

While the model assumes that egoistic voters cannot be intrinsically biased toward their groups, a fruitful way to interpret this hypothesis would be to expect an interaction effect between national bias and altruism. Somewhat counter-intuitively, my argument thus implies that altruistic people may be more, rather than less, opposed to immigration since they find it incompatible with their partiality toward compatriots. After all, unlike their non-altruistic counterparts, they are much more motivated by their ingroup preference. By the same token, however, I also expect both parochial and universal altruists to be more supportive of policies increasing immigration that explicitly benefit their country. At the same time, I expect all altruists to be more supportive of domestic redistribution and engage more in politics than egoists. These conventional social dilemmas (van Lange et al., 2014) pit self-interest against the collective good but do not have to pose conflict between national and global interests (Miller & Ali, 2014). Consequently, I also derive the following “placebo” hypothesis:

Hypothesis (1b: No altruist’s dilemma) Compared to egoists, people of either higher parochial and universal altruism are more likely to prefer policies increasing domestic redistribution.13

Does the relative prevalence of parochial rather than universal altruism imply that most electorates would never support increasing immigration? Although parochial altruism likely has various (relatively immutable) genetic and environmental determinants (Penner et al., 2005), its consequences may still be dependent on available policies and beliefs. My framework thus also implies that, independent of their self-interest and other concerns, most voters can be in principle supportive of increasing immigration or any other policy as long as it is believed to be helping others, but especially their compatriots. Consequently, I will also experimentally explore the counterfactual implications of the parochial altruism theory:14

Hypothesis (2a: Altruism) Independent of its personal impacts, the more positive the beliefs about the collective impacts of a certain immigration policy, the more supportive the voters are of that policy. Hypothesis (2b: Parochial altruism) Independent of its personal impacts, voters’ beliefs about the group impacts of a certain immigration policy are more predictive of their support than global impacts. Hypothesis (2c: National parochial altruism) Independent of its personal and global impacts, voters’ beliefs about the national impacts of a certain immigration policy are more predictive of their support than other group impacts.

Of course, it is commonly acknowledged that people tend to be more prosocial toward their ingroup. There is also a large literature suggesting that voter concerns about their nation is a significant factor in determining their policy preferences. But while the basic idea of parochial altruism is straightforward, the presented empirical implications challenge or otherwise inform some of the established theories (for a detailed review, see Kustov, 2019a). Most prominently, many social and political psychologists assert that, self-interest aside, people are motivated by their “social identity” (Tajfel & Turner, 1986). As a growing part of this scholarship, various aspects of national identity and “nationalism” are increasingly recognized to be important predictors of immigration attitudes (e.g., Kunovich, 2009). With the recent rise of populist and anti-immigration movements in Western democracies, it may almost seem a truism to suggest that those who are “nationalists” dislike immigration. Nonetheless, the term “nationalism” and its variations can be used to describe at least two psychological dimensions: (1) preference for one’s national group relative to other national groups (i.e., a form of group bias); and (2) preference for one’s national group relative to oneself (i.e., a form of altruism). As the “parochial altruism” accounts makes clear, these two dimensions are distinct and they interact with each other such that those who favor their compatriots over foreigners are expected to be especially anti-immigration when altruistic. Furthermore, theories of “sociotropic politics” suggest that people support or oppose a policy (increasing immigration) because they think it is good or bad not necessarily for themselves, but for others in their country (Hainmueller & Hopkins, 2014). Although this general expectation is consistent with the parochial altruism thesis, as of yet these accounts do not explore the role of individual predispositions or distinguish between self-interest and more or less parochial altruism as competing motivations (Kiewiet & Lewis-Beck,

2011). As a result it is simply unclear whether “sociotropic” voters should be expected to oppose a policy in their national interest when its personal (global) consequences are sufficiently worse or when these voters are themselves sufficiently self-interested (universally altruistic). In sum, while it might have been important to establish that people tend to choose immigration policies that they think are better for their countries, the theory of parochial altruism goes further by specifying when voters respond to national consequences and why some voters are more responsive to these consequences than others. At the same time, it is common to attribute anti-immigration attitudes to “taste-based” ethnoracial prejudice measured as ethnocentrism (e.g., Kinder & Kam, 2010) or negative stereotypes (e.g., Burns & Gimpel, 2000). According to these accounts, people form policy preferences based on their intrinsic motivation for derogation of ethnic outgroups regardless of perceived policy consequences (a part of eik term in the model). While my theory is not incompatible with the existence of such prejudice, H2a-H2c imply that the abundant ethnic biases observed in the literature are to a certain extent conditional on assumed policy consequences. Furthermore, as emphasized earlier, ethnic bias alone cannot explain why even racially egalitarian voters often do not support increasing immigration. Finally, as some scholars point out, people may decide to advance their perceived collective interests as a heuristic for their self-interest without being at all altruistic (Weeden & Kurzban, 2017). Accordingly, numerous studies have argued for a purely selfish account of the opposition to immigration (for a review, see Hainmueller & Hopkins, 2014). While my account similarly assumes that self-interest may be the main motivation (Î± Î² Î± Î² i i i i > > − − 1), it also specifies that it does not have to explain most of the variation in policy preferences, and thus prosocial motivations cannot be neglected in the analysis. Contrary to these approaches, I argue that opposition to immigration does not have to be rooted in just selfish concerns or unconditional derogation of foreigners. Instead, given widespread beliefs about its negative impacts, voters may resist migration because they care about others and favor their compatriots. Still, I specify the following alternative hypotheses:

Hypothesis (2d: Pure self-interest) Independent of personal impacts, beliefs about the collective impacts of a certain immigration policy are not related to how supportive voters are of that policy. Hypothesis (2e: Ethnic prejudice) Independent of any impacts, the lower the number or ethnic difference of immigrants due to a certain immigration policy, the more supportive the voters are of that policy.

Data and Research Design

How can one examine the relationship of parochial altruism and political preferences? The most straightforward way is to identify appropriate existing survey data and compare people’s political views and behaviors by their level of altruism and bias toward compatriots. In such a study using the Gallup World Poll and the US General Social Survey (see Kustov, 2019a), I do find that immigration and redistribution preferences are systematically related to individual differences in self-reported parochial altruism. First, I show that people of high altruism are much more likely to support domestic redistributive or race-targeted policies but are not more likely to favor immigration or foreign aid. Second, I demonstrate that respondents high in both altruism and national bias are the most opposed to immigration and respondents high in altruism but low national bias are the most supportive of immigration even controlling for other major covariates such as education, partisanship, religiosity, and racial bias. This initial evidence indicates that voters who favor their compatriots over foreigners are more, not less, opposed to immigration when they are altruistic. However, the procedure of correlating various endogenous survey responses—where respondents do not have an incentive to reveal their preferences—has its well-known downsides. The problem of “social desirability bias” may be especially pertinent in the study of altruism. As a result, scholars have increasingly incorporated various incentivized monetary games in their political surveys to alleviate these concerns (for a discussion, see Gilens & Thal, 2018). Moreover, since even revealed altruistic preferences can be endogenous to a variety of other predispositions, it is also important to consider causal evidence for my argument. To test my argument against existing accounts in a compelling way, I thus conduct an original population-based survey with embedded incentivized economic games and choice experiments. I concentrate the following empirical analysis on the United Kingdom—an advanced democracy and a top immigration destination whose migration policies are crucial for international freedom of movement. While the British public has always been divided on the issue, immigration attitudes have become increasingly negative after the opening of its labor market to Eastern Europe in 2004. This widespread opposition to immigration has been further linked to the relative success of the euroskeptic UK Independence Party (UKIP) and the “Leave” vote in the 2016 EU referendum (Goodwin & Milazzo, 2017). Overall, despite some peculiarities related to country’s electoral politics and its institutions, the political conflict over immigration in Britain is representative of other advanced democracies.

All the analyses presented below are based on the quasi-representative survey of 1963 English15 citizens administered online by Qualtrics in May 2018.16 To ensure cross-national generalizability, I also complement these data with a pilot sample of 604 native-born US respondents recruited via Amazon MTurk in January 2017.17 The main survey was comprised of the following five consecutive blocks: immigration attitudes, conjoint experiment, other attitudes, demographics, and incentivized economic game.18

Below, I first provide observational evidence for my argument by revealing parochial altruism via an incentivized charity game with various recipients and then relating these preferences to political behavior (“Study 1”). I then experimentally test broader counterfactual implications of my theory by estimating the effect of perceived immigration consequences in a conjoint analysis of policy alternatives (“Study 2”). Importantly, both studies examine the role of parochial altruistic motivations in shaping immigration policy preferences, but they approach the question in a distinct but complementary way. In particular, while Study 1 directly relates individual-level measures of immigration attitudes and (revealed) altruistic predispositions, Study 2 infers the role of altruism from how the relevant policy consequences (such as related to national interest) affect immigration preferences in the aggregate.

Study 1: Why Do Altruistic Voters Resist Immigration?

Observational Design: Elicitation of Parochial Altruism

In creating my original measure of parochial altruism, I build on the growing literature that uses economic games to reveal individual preferences. In particular, I rely on a variant of the “dictator game” where subjects are given real money and then they have to decide whether to share it with others. Although giving behavior in these games can be subject to various interpretations (e.g., “warm glow”), these measures have been shown to be systematically related to economic and political behavior as-if indicative of altruism (Dawes et al., 2011; Fehr & Schmidt, 2006). To better adapt this behavioral measure of altruism for an online survey setting, I follow the recent literature and use participation in a real-stake lottery with an option of charitable giving as the main incentive (Bechtel et al., 2019; Falk et al., 2016).19 To reveal more or less parochial altruism in particular, I introduce a modified version of such charity dictator game (“altruist’s dilemma game”) which exploits the allocation decisions between respondents’ own account and various national-oriented or global-oriented charities.

Specifically, all respondents were informed that they could either decide to keep or donate (a part of) the £ 100 prize if selected. They then had to choose potential recipient(s) among a randomized list of six well-known national and global charities (matched by specialization), a different charity of their choice, or their personal account.20 In other words, respondents were incentivized to choose (a combination of) their self-, national, and global interests and thus reveal their altruistic preferences in the absence of any reputational or other confounding concerns. While there can be various ways to utilize the data, the most straightforward way to operationalize parochial altruism is to divide the sample into three “types” based on respondents’ revealed altruistic priorities.21

Analysis and Results

In line with the theoretical model (Figure 2), Figure B1 visualizes the resulting distribution of revealed preferences. As it may be expected, the majority (57%) of respondents (“egoists”) decided to keep the whole £ 100 prize. At the same time, 30% (“parochial altruists”) made a primary contribution to one of the national-oriented charities, while 13% (“universal altruists”) mostly made a contribution to global-oriented charities (including 4% who contributed to national and global organizations equally). Furthermore, approximately the same amount of donations was present across most prominent subgroups differentiated by gender, age, income, education, ideology, religiosity, and ethnocentrism. Overall, while genuine altruism was abundant, much more respondents opted for advancing their national rather than global interest (also see Buntaine & Prather, 2018). Quite importantly, these differences cannot be explained by the greater credibility of domestic charities since, for instance, more than twice as many people chose to commit to the Red Cross projects at home rather than abroad. Does this joint variation of altruistic and parochial preferences relate to important political choices? First, I test the “placebo” hypothesis by looking at the relationship of altruism and the conventional social dilemmas of political participation and redistribution. As can be seen from Figure 3 (and Table B2), both parochial and universal altruists are indeed more likely to participate in politics and support pro-poor welfare spending than egoists, which is overall consistent with the previous findings (Dawes et al., 2011; Gilens & Thal, 2018). As I argued above, however, the distinction between altruist types should become crucial when policies pose a tradeoff between national and global interests. When it comes to increasing immigration, which is predominantly perceived as detrimental to one’s country despite its benefits for the world,22

Figure 3. Political behavior and policy attitudes by parochial altruism. Note. Based on the England Qualtrics sample. Revealed preference types are based on participants’ contributions in the incentivized charity raffle. The bars represent 95% CI. For variable descriptions, see Online Appendix.

parochial altruists—the majority of prosocial respondents—are as opposed to it as those who are not altruistic. Only universal altruists—the minority of prosocial respondents who also take into account global impacts—are substantially (at least 0.5-0.8 standard deviations) more pro-immigration than other groups. While people are often uncertain about the personal effects of immigration,23 their reliance on perceived social consequences is clearly dependent on their revealed altruistic preferences. These differences in preference type persist even after controlling for demographics and other major covariates (see Table B2). In fact, the coefficient size of universal altruism is comparable to or even larger than that of education and ethnocentrism—the strongest predictors in the literature. For strikingly similar US results, suggesting that the consequential altruistic divide in the electorate is a general phenomenon, see Figure B2. Nonetheless, as expected, parochial altruism can also be related to increased support for immigration under certain conditions. While only a small percentage (12%) of voters are willing to relax existing restrictions in principle regardless of its national impacts, most people (58%) also say that they are willing to do it at least for those immigrants who are expected to contribute to the national economy.24 As can be seen from Figure 4, such policies also yield a relatively greater support from both parochial and universal altruists (the difference between altruists types for these policies compared to others is smaller at p <.02). While the level of nation-regarding support for immigration among parochial altruists is still not as high as among universal altruists, it is substantively more important given the much greater size of the former group.

Figure 4. Immigration support by national impact and parochial altruism. Note. Based on the England Qualtrics sample. Revealed preference types are based on participants’ contributions in the incentivized charity raffle. The bars represent 95% CI. The outcomes indicate the share of voters who support relaxing immigration restrictions regardless of national impact (left) or conditional on positive national impact (right). For variable descriptions, see Online Appendix.

Discussion

Contrary to the conventional wisdom that altruism is inextricably linked to immigration support, I provide new revealed preference evidence demonstrating systematic variation in this link related to the scope of altruism and the perceived national impacts. Put differently, I show that compassionate individuals are rarely impartial in their compassion and, when it comes to immigration policy, they are more likely to advance the perceived interest of their own national group even if it comes at the expense of a more general collective good. In doing so, I also provide a new way of measuring revealed nationalist (vs. cosmopolitan) preferences as the bias (or its absence) in altruistic behavior. Given the low variation in self-reported attitudes (i.e., voters almost unanimously report strong national and weak supranational attachments), revealed measures can fruitfully complement the existing empirical literature on the subject (e.g., Bayram, 2015; Herrmann, 2017). Accordingly, the England Qualtrics sample indicates that individual donations to domestic versus global charities are predictive of anti-immigration attitudes and other political behaviors even after controlling for conventional national and global identity items (not shown).

While the analysis does not show that parochial altruists are more anti-immigration than egoists, it is still quite striking that people who incur personal costs to help others have similar migration preferences to those who merely wish to advance their self-interest. This is especially notable considering that, similar to those who donate globally, those who donate domestically are genuine altruists who are more likely to participate in politics and support domestic redistribution than those who do not donate. The fact that egoists agree on immigration with parochial altruists in some cases but not others further casts doubt on the view that voters use information about group impacts as a mere heuristic to self-interest. By the same token, this suggests that widespread sociotropic concerns about immigration impacts are at least in part driven by (parochial) altruistic motivations. Given the contextual limitations of the revelation procedure, however, it may be reasonable to interpret the decision of donating most or all of the (hypothetical) winnings as a marginal difference in pure altruistic preferences. For instance, it is still possible that many of those who decided to keep the winnings have significant parochial (and some perhaps even universal) altruistic preferences that might have been revealed under a more elaborate repeated procedure. Consequently, while my analysis focused on the differences between revealed voter types, it is possible they captured only a part of the underlying continuum. Finally, beliefs about policy impacts can be potentially endogenous to parochial altruism due to motivated reasoning (Herrmann, 2017). Consistent with this idea, the England Qualtrics sample indicates that parochial (universal) altruists are more (less) likely to view increasing immigration as globally and nationally beneficial compared to egoists (not shown). As a result, it is essential to consider experimental evidence for my argument.

Study 2: When Would Most Voters Support Immigration?

How can the causal connection between altruistic motivations and immigration policy preferences be uncovered? Ideally, this requires the experimental manipulation of intrinsic prosocial motivations cognitively independent from the outcome of interest, which is difficult outside of the lab (for an exception, see Lacetera et al., 2012). While the literature also provides a variety of more or less unobtrusive ways to prime altruism (Uhlmann et al., 2009), ingroup bias (Hertel & Kerr, 2001) and national identity (Hassin et al., 2007), there has been growing criticism regarding the robustness of such priming results (Kahneman, 2012).

Instead, I propose to estimate the relative effect of altruistic motivations on immigration preferences (average Î± and Î² parameters) by manipulating the perceived policy consequences (I G T ik jk k;;) in a conjoint survey experiment in which respondents choose between a set of hypothetical immigration policies. In doing so, I employ an established method for eliciting preferences underlying trade attitudes (Baron, 2012; Mutz & Kim, 2017), complement it with a multidimensional choice experiment technique (Hainmueller et al., 2014), and then apply it to the issue of immigration policy preferences as opposed to individual admission preferences explored in the earlier conjoint research (e.g., Hainmueller & Hopkins, 2015).

Experimental Design: Conjoint Analysis of Policy Choice

As of now, voters overwhelmingly oppose policies increasing immigration. According to the logic of parochial altruism (H2a-H2c), however, they would have been more supportive of increased immigration if (all else equal) they had perceived it to be more nationally beneficial. Quite equivalently, people should be more supportive of alternative pro-immigration policies that are explicitly designed to be more beneficial to their country. But what about a similar proposal that might also substantially undermine personal or global prosperity? Are voters actually willing to tradeoff the interest of their country against their own self-interest or the world as a whole? What about other important non-economic considerations? To answer these questions and uncover the influence of altruistic motivations, I use an original conjoint experiment of hypothetical immigration policy choice. Specifically, in my conjoint experiment I ask respondents to choose their preferred immigration policy from five policy pairs with randomized attributes (so that each respondent sequentially selects 5 out of 10 distinct policies). All policy attributes were randomly selected from discrete, pre-defined levels indicating heterogeneous implications for the number of immigrants from different regions, as well as (economic) self-, local, national, and global consequences of various magnitude. The main outcome question optimally forces respondents to choose one of the policies in a given pair (see Hainmueller et al., 2014). Overall, the conjoint has the following six attributes and the respective randomized levels (for an example, see Figure 5):25

Number of immigrants: allow almost none, allow some, allow many, allow almost all 2. Sending region: Western Europe, Eastern Europe, Asia, Middle East, Africa

Figure 5. An example of a conjoint policy choice task (UK).

Your household wealth: decreased by 5% to 6%, decreased by 1% to 2%, no change, increased by 1% to 2%, increased by 5% to 6% 4. Your city or town’s wealth: decreased by 5% to 6%, decreased by 1% to 2%, no change, increased by 1% to 2%, increased by 5% to 6% 5. British wealth: decreased by 5% to 6%, decreased by 1% to 2%, no change, increased by 1% to 2%, increased by 5% to 6% 6. Global wealth: decreased by 5% to 6%, decreased by 1% to 2%, no change, increased by 1% to 2%, increased by 5% to 6%

One of the main advantages of such design is its ability to disentangle the causal connection between immigration attitudes and beliefs about policy sequences, which is generally hard to do in real-world politics due to motivated reasoning (e.g., Kahan, 2016). Given that all the policy attributes are randomized simultaneously and their effects are measured on the same scale, the design allows estimating and comparing the elasticity of (counterfactual) immigration preferences to various personal and collective interests. In turn, this enables examining the relative importance of (potentially correlated) egoistic and altruistic motivations that underlie immigration attitudes.26 The addition of “immigrant number” and their “sending region” attributes further helps accounting for other potential concerns in voters’ decision-making. Most prominently, it is possible that—due to their ethnic preferences or some other reason27—respondents may decide to reject policies that increase immigration even if, as explicitly stated, they are economically beneficial for everyone.

To test my hypotheses regarding the relative importance of various motivations (H2a-H2e), I follow the empirical approach developed by Hainmueller et al. (2014) and estimate average marginal component effects (AMCEs) of various attributes on policy choice using simple linear regression with robust standard errors clustered by respondent. The AMCE represents the average difference in the probability of being chosen when comparing two different attribute values (e.g., a policy proposal of immigration from Western Europe versus Africa) where the average is taken over all other possible attribute combinations. Finally, to examine heterogeneous effects by altruist type, I compare the respondents who make a donation to charity to those who keep their money in the economic game from the previous study (see Online Appendix). All the main and alternative conjoint hypotheses (including heterogeneous effects), as well as the analysis plan were pre-registered in the Evidence in Governance and Politics (EGAP) repository (#20180529AB).

Analysis and Results

Selfish and Altruistic Incentives. Figure 6 visualizes the experimental results based on the estimates from the benchmark OLS regression (Table B3, 1) of policy choice on various economic consequences for all English participants in a full sample. Given that there are almost no substantively significant interactions between treatments (see Table B4), most of the effects discussed below can be considered independent of each other as stipulated in the theoretical model. As may be expected by political economy theories, respondents’ preferences are highly responsive to personal policy consequences (when a policy explicitly harms or benefits themselves and their household). Specifically, the difference in the probability of policy selection between the worst and best possible self-interest outcome is about 28 2 ± % (see Figure 7). While a policy’s impact on household wealth is by far the strongest single predictor of respondent support, the results also indicate that people care about collective consequences in addition to personal consequences. In fact, contrary to theories of pure egoism, the combined independent effect of local, national, and global interests is even greater than the effect of self-interest alone (44 3 ± %, also see Figure 8). Put differently, an average voter may even decide to incur a cost (or forgo a benefit) to herself if some political alternative is clearly beneficial (detrimental) to everyone, despite the predominance of her selfish motivations.

Figure 6. Effects of immigration policy consequences on selection. Note. The plot shows the effect estimates of the randomly assigned policy consequences on the probability of being selected. Estimates are based on the baseline OLS model of the England Qualtrics sample (Table B4). Bars represent 95% CIs. For variable descriptions, see Online Appendix.

Figure 7. Effects of best versus worst immigration policy consequences on selection. Note. The plot shows the effect estimates of the randomly assigned best versus worst policy consequences on the probability of being selected. Estimates are based on the baseline OLS model of the England Qualtrics sample (Table B3). Longer (shorter) bars represent 95% (84%) CIs. For variable descriptions, see Online Appendix.

Figure 8. Effects of personal and social consequences on selection by revealed altruism. Note. The plot shows the effect estimates of the randomly assigned best versus worst personal and collective consequences on the probability of being selected by respondents’ decisions to donate their money to charity in the pre-treatment economic game. Collective interests are calculated as a combination of nonpersonal interest treatments. Estimates are based on the OLS model of the England Qualtrics sample similar to the baseline (see Table B3). Longer (shorter) bars represent 95% (84%) CIs. For variable descriptions, see Online Appendix.

To better compare the effects of various collective consequences, Figure 7 also visualizes the experimental estimates of the difference in the best and the worst possible policy outcomes (5-6% increase vs. 5-6% decrease).28 In line with H2b, people significantly prioritize parochial to global relative gain, despite the fact the latter is much greater in absolute terms.29 Furthermore, when it comes to various parochial impacts (H2c), the probability difference in selection for national interest is expectedly the most influential (19 2 ± % points). While voters also greatly care about their local community, the average effects are less prominent (14 2 ± %). Consequently, contrary to the common understanding that people necessarily prioritize groups that are spatially closer, national membership does seem to dominate over local allegiances. The results for global interest are quite striking as well. While it is not particularly surprising that its independent effects are rather weak (11 2 ± %), their very existence is noteworthy. Conventional ideas of omnipresent tribalism notwithstanding, the evidence thus suggests that voters are willing to take at least some global concerns into account, even when choosing national policy (also see Buchan et al., 2011).30

Finally, if greater sensitivity to collective interests revealed in the conjoint task is indeed indicative of genuine altruism rather than a group heuristic, we should also expect it to be more pronounced among the altruists independently revealed in the charity raffle game from the previous study. As indicated by

Figure 9. Effects of immigration policy attributes on selection. Note. The plot shows the effect estimates of the randomly assigned policy characteristics on the probability of being selected. The treatments represent the categorical measures for immigrant numbers and the sending region. Estimates are based on the baseline OLS model of the England Qualtrics sample (Table B3). Bars represent 95% CIs.

Figure 8, those who decide to donate to charity are less responsive to changes in household wealth and suggestively more responsive to a combination of collective (local, national, and global) interests. The comparison of coefficient differences between the two indicates that revealed altruists are more likely to prioritize collective over personal interests in their policy choice (at p <.01). While these differences are not large, they are consistent with the interpretation of respondents’ donation decisions in the incentivized task as indicative of only marginally greater altruism.31

Immigration Number and Sending Region. Nonetheless, it is possible that immigration preferences are also determined by people’s categorical rejection of certain ethnic outgroups. According to Figure 9, policy preferences are indeed responsive to the number of immigrants and their sending region even when policy consequences are taken into account. In particular, the independent

negative effect of increasing immigration (“allow almost all” or “allow many”) compared to limiting immigration (“allow some” or “allow almost none”) is quite considerable (12 15 3 − ± %). Furthermore, there is also a negative effect of non-European immigration (6 7 3 − ± %). How should we interpret the magnitude of the observed effects? On the one hand, the estimates are indeed substantively significant. For instance, a policy that allow almost all immigration is supported as much as a policy that causes a 5% to 6% decrease in national wealth. At the same time, a policy that approaches non-European immigration is supported as much as a policy that causes 1% to 2% decrease in national wealth. The more detailed subgroup analysis further suggests that high-immigration policies are especially pronounced among people with higher ethnocentrism (Figure B3). In line with this, people with strong ethnic preferences are also slightly less responsive to both personal and social incentives (see Figures B4). In other words, some are willing to tradeoff their personal and even national material well-being for living in a more ethnically homogeneous community (also see Kaufmann, 2019). On the other hand, the effects of immigrants’ number and their sending region are much less influential than various policy consequences as a whole for most voters. In particular, the former explains less than one-third as much variance in policy selection as the latter (2% vs. 7%). Albeit less so, the same holds true even for ethnocentric voters. Although the particular numbers may be specific to the experimental design used in the study, it is likely that the influence of ethnic prejudice and related preferences is far less pronounced than that of self-regarding and other-regarding concerns.32 In other words, the categorical rejection of immigration due to taste-based ethnic prejudice and related preferences is present among some respondents, but far from pervasive.33

In sum, even after accounting for self-interest and non-economic concerns, immigration preferences are highly responsive to perceived national consequences but not so much to local and even less so to global consequences. This, in turn, is indicative of widespread national parochial altruism. Conditional on those consequences, however, there are only moderate differences in policy support related to the number of immigrants and their origin (which are driven by ethnic preferences). To check the robustness of these results, I conduct a number of additional empirical tests with no change in the substantive findings (see Online Appendix).

Discussion

Overall, the results of my theoretically-driven choice experiment suggest that voters’ immigration preferences are indeed highly responsive to their perceived national interest in addition to their self-interest. Furthermore, conditional on these perceived interests, immigration preferences are only moderately responsive to the number of immigrants or their sending region. According to the estimated probability of policy selection based on the benchmark model (not shown), all their concerns notwithstanding, most voters could thus support increasing immigration from non-European countries if they believe it would benefit themselves and their compatriots. Consequently, despite current political realities, the prevailing sentiment of “my country first” could be in principle employed to increase immigration levels. As indicated by the similarity between the UK and the US results, this pattern of national parochial altruism extends across advanced democracies. In the real-world, the role of selfish versus altruistic motivations is often confounded by the fact that people are more uncertain about personal than collective consequences of various political alternatives. As a result, when voters respond positively to national economic growth, it is unclear the extent to which they use this as a group heuristic for their self-interest. As the methodological innovation introduced here, the conjoint analysis of policy choice can differentiate between these motivations, so that the independent effect of national interest can be considered an indication of parochial altruistic preferences at work. At the same time, this evidences contributes to a better understanding of the role of ethnoracial preferences in immigration politics. Of course, the very existence of ethnic bias in the results is not particularly surprising given the abundant evidence on the cultural and demographic drivers of anti-immigration attitudes (Hainmueller & Hopkins, 2014). Nonetheless, one may argue that when people are asked about immigration in a survey they may have unobserved expectations about policy consequences. Contrary to that, the conjoint setup employed here explicitly specifies both these consequences and the immigrant origin using random assignment. The conjoint can thus reveal biases regarding different origins while controlling for policy consequences. As a result, these estimates are arguably more precise at capturing the scope of ethnic prejudice and related policy-specific concerns in determining public attitudes on immigration. Of course, the study has a few notable limitations including those of survey research and hypothetical choice experiments more generally. First, in focusing on changing immigrant admission numbers and the related (material) consequences, the conjoint does not address important policy

preferences with regard to immigrant integration and rights. For instance, it is likely that voters may also prefer to welcome more immigrants who are willing to assimilate (Hainmueller & Hopkins, 2014) or who would have fewer citizenship rights (Ruhs, 2013). If these or other “cultural” concerns about immigration were indeed the most dominant, however, one would expect voters to have been significantly more responsive to immigrant region and their number in the conjoint task. Second, it is possible that some respondents may have not “believed” certain policy profiles, especially when they contradicted their strong priors on immigration (e.g., an increase in non-European immigration leading to more national wealth). According to their own open-ended responses, however, most participants interpreted the task in the way intended by the author and accepted hypothetical policy impacts as given. This important concern about design validity is further alleviated by the fact that removing atypical policy profiles or conditioning on prior beliefs about immigration impact does not change the main results. Third, one can reasonably argue that, even when properly understood, the conjoint task with a number of counterfactual policy attributes including such abstract concepts as “global wealth” may be too cognitively demanding for some respondents. Nonetheless, when asked to explain their decisions in a open-ended question, most respondents expectedly alluded to some combination of their personal, ethnic, national, or global concerns. Overall, in as much as the relevant decision-making alternatives are adequately specified in the conjoint design, the results should be predictive of real-world political behavior (for a relevant methodological discussion, see Hainmueller et al., 2015).

Conclusion

While severe government regulations on migration adopted over the last century have significant human costs, most voters in high-income countries do not want to relax these restrictions. This dissonance is peculiar because there can be a number of alternative Pareto-improving policies that benefit everyone. Given the recent humanitarian crisis of displaced persons and the related need for novel democratic solutions, it is especially imperative to better understand the motivational basis of widespread anti-immigration preferences. As of now, it is common to explain these public preferences by alluding to voters’ selfishness and ethnic animus. To understand why even educated and liberal voters support welfare-diminishing restrictions, however, scholars should move beyond the accounts of inherently “self-interested” or “prejudiced” voters. To that end, my research highlights the unlikely role of genuine

altruism in popular resistance to immigration. In particular, I argue that this opposition stems from the well-documented fact that most people, included those who are educated and liberal, care about others but also prioritize their compatriots. Given the persistent public perceptions about the national risks of mass immigration, widespread altruistic motivations are actually not conducive to popular support of international mobility. I then formally contrast my theory to competing accounts, derive hypotheses about voter preferences, and specify appropriate empirical methods to test it. Building on existing data, I provide initial self-reported evidence that altruistic voters are generally anti-immigration and often much more so than those who are not altruistic. To address the social desirability and endogeneity concerns, I then develop, pre-register, and run an original populationbased UK study. First, to elicit altruism and its link to migration preferences, I include an incentivized charity raffle. Second, to get at the causal effects of people’s perceptions of immigration impacts, I include a conjoint experiment of hypothetical policy choice. Overall, I find that most voters, and especially those who are altruistic, can only support increasing immigration when they truly believe it benefits their compatriots. At the same time, my results also indicate that those who unconditionally prefer to stop migration at any cost are in minority. Why would people believe that increasing immigration is in their personal and national interests? While current restrictive preferences may be driven by widespread (mis)perceptions about immigration’s negative impact, there is little evidence that immigration is actually harmful to receiving states and their citizens (for a review, see Blau & Mackie, 2016). Nonetheless, it is also the case that most of the unrealized gains of immigration would go to potential immigrants and not to the voters who have to decide on the issue, at least in the short term (Clemens, 2011). Consequently, it may be hard to convince people that more immigration is good for most citizens. Consistent with this idea, most attempts to change people’s mind on immigration by providing facts have not been successful (Hopkins et al., 2019; Kustov et al., 2019). The often overlooked alternative, however, is to change the policy environment itself. While it is common for policymakers to motivate various immigration restrictions by alluding to national interest, it may also be possible to increase immigration by designing more optimal policy solutions that explicitly and straightforwardly benefit average citizens and migrants alike (e.g., see Clemens, 2015). The results here demonstrate that most voters should be willing to compromise their current anti-immigration sentiments and support policies increasing immigration when they are confident in their favorable national consequences.

Stepping aside from the immigration debate, my research provides new evidence on what motivates people in politics. Specifically, I demonstrate many voters respond to altruistic incentives even when it is against their own interest. In doing so, however, they prioritize the costs and benefits to their compatriots, although they have no obligation to do so as private benefactors. The related individual differences in altruistic behavior systematically predict people’s political choices in addition to their other characteristics and identities. As a result, the idea that genuine altruistic motivations may often be in conflict appears can be useful in explaining political behavior across a variety of outcomes and contexts. The proposed distinction between altruist types should be especially important in political as opposed to economic behavior since people are often uncertain about personal impacts of various political alternatives and their choices are rarely personally consequential (Schnellenbach & Schubert, 2015). This division can further complement a prominent—but often less analytically useful—juxtaposition of economic and cultural behavioral causes. For instance, the current debate about the rise of populism that mostly focuses on economic anxiety and cultural backlash as exhaustive explanations (e.g., Norris & Inglehart, 2019) may be missing an important preference variation in the electorate (also see Kustov, 2019b, 2020). In distinguishing between more or less parochial altruism, this paper also informs the long-standing debate on how individuals prioritize their identities and interests in a competitive political environment (Druckman & Lupia, 2016; Wong, 2010). For instance, while my results are not consistent with the accounts of pure or even predominant self-interest in political preferences (e.g., Weeden & Kurzban, 2017), they also go against the widespread idea that self-interest matters only in exceptional circumstances (Sears & Funk, 1990). Furthermore, contrary to the conventional wisdom that “all politics is local,” the fact that voters care more about their national rather than local interests is in line with the growing evidence of greater nationalization of politics (Hopkins, 2018). Using revealed preference evidence rather than self-reported attitudes, this study thus also corroborates the idea of nationality as a defining political category in the modern world (Schildkraut, 2014). Finally, if anti-immigration attitudes are related to the genuine willingness to help compatriots more than foreigners, the prominent distinction between ingroup love and outgroup hate (Brewer, 1999)—or “benign” patriotism and “malignant” nationalism—may be irrelevant when global and national interests are not aligned. In this respect, patriotism normally emphasizes the pro-social part and nationalism emphasizes the antisocial part of the exact same motivation. As opposed to existing sociotropic accounts, however, the study also clarifies that voters would generally not choose a policy that is beneficial

to their compatriots at a massive expense of themselves or everyone else. In the end, my argument implies that both positive and negative consequences of group identity can be amplified by altruism. Over the last years, both the Brexit referendum in the United Kingdom and the election of Donald Trump in the United States have often been described as exemplifying the growing electoral divide between “nationalists” and “globalists” or “communitarians” and “cosmopolitans” (e.g., Haidt, 2016; Hooghe & Marks, 2018; Norris & Inglehart, 2019). As my account makes clear, however, altruism is exactly what gives motivational power to these conflicting identities. Since most people today say they love their country and care about the world (e.g., see Bayram, 2015), it may be more instructive to ask whether (and to what extent) voters are actually willing to sacrifice their own well-being for the sake of national or global interest. Consequently, the respective individual differences in (parochial) altruism can provide the psychological basis for an increasingly important political cleavage on globalization across the world. Of course, the evidence presented in this paper also opens up a number of questions for future research. For instance, why are some people and societies more parochial and altruistic than others? While parochial altruism as an individual predisposition is assumed to be rather stable and universal, the way it manifests itself in politics can be systematically affected by institutions and other contextual factors. For instance, the dominance of national parochial altruism relative to other types of ethnic loyalties is likely limited to modern nation-states with minimal democratic institutions. Consequently, future research can benefit from examining the contextual variation of parochial altruism (see Hruschka & Henrich, 2013), as well as the factors that constrain or encourage these motivations across time and space. All in all, regardless of whether one believes that (national) parochial altruism or any other predisposition is normatively desirable, it is important to agree on the empirical reality of what motivates people in politics. After all, widespread parochialism implies that, under certain conditions, sincere altruistic motivations can not only promote cooperation such as solving collective action problems, but can also exacerbate conflict such as in the case of immigration.34 In this respect, the ultimate question is what institutions can help societies reap the benefits of parochial altruism but also minimize its domestic and global costs.

Authors’ Note

An earlier version of this manuscript received the Best Paper Award at the 2019 Annual Meeting of the American Political Science Association (honorable mention, Migration and Citizenship section).

Acknowledgments

I would like to thank my colleagues who have read and commented on previous drafts of this article. I am especially grateful to Rafaela Dancygier, Samuel Diaz, Marty Gilens, Tali Mendelberg, and Grigo Pop-Eleches for their helpful suggestions. For their useful comments on the previous versions of the larger project on the politics of parochial altruism, I would also like to thank Chris Achen, Meir Alkon, Jon Baron, Winston Chou, Sharan Grewal, Ronald Inglehart, Becky Johnson, Amanda Kennard, Ben Kantack, Dela Kpo, David Laitin, Egor Lazarev, Diana Mutz, Giuliana Pardelli, Tanika Raychaudhuri, Jake Shapiro, Nick Valentino, and Ali Valenzuela. Different parts of the project were also presented at APSA, CESS, HEWG, MPSA, SPSA, Reppy, and TPBW meetings, as well as Princeton research seminars. All errors and omissions are the sole responsibility of the author. This study was approved under Princeton IRB #7888. The pre-analysis plan for the conjoint experiment was archived in the Evidence in Governance and Politics (EGAP) repository (#20180529AB).

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Author is grateful to the Mamdouha S. Bobst Center for Peace and Justice, as well as Princeton Research in Experimental Social Science (PRESS) for their financial support.

ORCID iD

Alexander Kustov https://orcid.org/0000-0001-9361-8739

Supplemental Material

Supplemental material for this article is available online.

Notes

See https://yougov.co.uk/news/2018/05/11/yougov-data-reveals-what-europeans-think-are-most-/YouGov and https://news.gallup.com/poll/237389/immigration-surges-top-important-problem-list.aspxGallup reports. 2. Most immigration policies are laws aimed at restricting admission or rights of non-citizens (Ruhs, 2013). 3. For the comprehensive review of economic estimates, see Pritchett (2006); Clemens (2011).
For the most recent global evidence, see International Organization for Migration (2015). 5. For an overview, see Dawes et al. (2011); Gilens & Thal (2018). 6. While parochial altruism—similar to any other predisposition—is a continuous individual difference, the evidence based on the incentivized donation decisions described below reveals that approximately 30% of the UK respondents can be classified as “parochial altruists,” 9%—as “universal altruists,” and 57%—as “egoists.” 7. For an alternative, albeit related, formalization of group bias and other-regarding preferences in a unified framework, see Robalo et al. (2017). For a review of general altruism models, see Rotemberg (2014). 8. Note that I is already assumed to include potential individual return from group and global good. For a more explicit modeling of this process in a public good game setting, see Böhm et al. (2014). 9. In light of the persistence evidence of people’s scope insensitivity (Dickert et al., 2015), I model preferences in relative rather than absolute terms. Furthermore, given the complexity of political information environment (Druckman, 2014), the model does not differentiate between actual effects and individual beliefs, however correct, regarding these effects. Finally, while people may choose to join or exit certain groups, the model only considers (ethnic) categories that are as-if exogenous and immutable in the short term (Chandra, 2006). 10. This is not to say that genuinely impartial altruism of high personal cost does not exist or is very rare. For the accounts of rescuers during the Holocaust, for instance, see Oliner & Oliner (1988). For the discussion of the Effective Altruism movement, see Singer (2015). More generally, there is a substantial evidence for the existence of cosmopolitan identities and more or less universal altruistic preferences (Buchan et al., 2011). 11. While some scholars assume that outgroup hostility is an inherent part of “parochial altruism,” it is more fruitful to separate these distinct phenomena (Rusch, 2016). For the purposes of this study, the model thus assumes that individuals cannot attach a negative value to others’ well-being (other than in the eik term). 12. More formally and generally, parochial altruists (or individuals with greater Î²) are less likely to choose alternative k (e.g., increasing immigration) given that their group payoff from it G jk is assumed to be negative regardless of Tk. At the same time, universal altruists (or individuals with smaller Î± and Î²) are more likely to choose k given that the total payoff from it Tk is assumed to be positive regardless of G jk. 13. More formally and generally, altruists (or individuals with smaller Î±) are more likely to choose alternative k given that their group and total payoffs from it

G jk and Tk are assumed to be positive. 14. The first hypothesis naturally follows from the assumption that G j (group payoff) is additively separable from Ii (personal payoff) and that Î±i (personal payoff weights) <1. The second hypothesis follows from the assumption that

T (total payoff) is additively separable from G j (group payoff) and that 0 < Î²i

(group payoff weights) < − − 1 Î± Î² i i. Independent from the model, the third hypothesis is based on the evidence of relatively high salience of national citizenship in modern democracies. 15. The study is focused on England rather than the UK as whole to avoid potential confounding of British and other strong national identities in the state (McCrone, 2002). 16. As specified in the pre-analysis, this diverse national sample targeted to match census demographics was obtained from the initial pool of 2050 respondents after accounting for response quality (attention check and survey completion) and excluding non-citizens. The inclusion of all respondents in the analysis, however, does not affect the results (not shown). The employed Qualtrics panel was representative of the population across most important demographic and political characteristics (see Table B1 for summary statistics). 17. The sample was obtained from the initial 647 respondents after accounting for response quality (attention checks) and excluding non-citizens. 18. In the pilot sample, the participants were randomized to participate in the economic game either in the very beginning or the end of the survey without any differences in the relevant predictors or outcomes. 19. As in the case dictator giving, altruism is only a motive of charitable giving (Bekkers & Wiepking, 2010). 20. The list is made of the following charities: British Red Cross; International Committee of the Red Cross; National Society for the Prevention of Cruelty to Children; Save the Children International; Shelter, the housing and homelessness charity; International Rescue Committee. For vignette and details, see Online Appendix. 21. Those who made a greater or equal (lesser) donation to global than national charities are coded as universal (parochial) altruist type. The results are robust to the continuous operationalization of parochial altruism. 22. Their disagreements notwithstanding, voters are more pessimistic about national than global consequences of immigration. According to the England Qualtrics sample, while a plurality of respondents acknowledge that increasing immigration is more positive than negative for the world (33% vs. 29%), most believe it is more negative than positive for their country (47% vs. 32%). Due to the endogeneity of these perceptions (see below), I do not use them in my observational analysis. For variable descriptions, see Online Appendix. 23. According to the England Qualtrics sample, many more respondents are indeed uncertain about personal than national or global effects of increasing immigration (46% vs. 25% or 38%). 24. According to the England Qualtrics sample, approximately the same number of voters would also support a policy significantly increasing immigration if it is specifically designed to benefit average British citizens through greater selection and taxation of immigrants (not shown). 25. The sequence of attributes 3-6 was randomized. The introductory vignette is given in Online Appendix. The relatively small number of attributes and levels

(4 5 5 5 5 5 × × × × ×) ensures that respondents are able to adequately assess presented policy counterfactuals. While some effect magnitudes used in the study are arguably more realistic than others, there is no indication that respondents’ perceptions of policy feasibility affected their choices given their own explanations in a follow-up open-ended question (also see Discussion). 26. The design deliberately does not pose explicit tradeoffs between one’s ingroup and outgroup to avoid confounding altruism with a variety of group-based psychological mechanisms related to reciprocity and other social norms (e.g., Yamagishi & Mifune, 2008). 27. For instance, people may also want to limit immigration from certain regions due to perceived security concerns or their preference for more culturally homogeneous or integrated communities. 28. Despite the discreteness of all treatment policy attributes, this transformation can be justified given the actual linearity of most observed effects in 6. 29. While unlikely due to scope insensitivity (Dickert et al., 2015), it is possible that people would be more sensitive to global gains if they are presented in absolute terms. 30. Another pattern that emerges from Figure 6 is the consistent dynamic of both self-regarding and altruistic loss aversion (e.g., Mercer, 2005). While outside the scope of my theory, such tendency may further exacerbate the political implications of parochial altruism and the tragedy of global commons: a 1-2% decrease in national wealth trumps the positive effect of a 5-6% increase in global wealth, but a 1-2% decrease in global wealth does not trump a 1-2% increase in national wealth. 31. This is the exact comparison that is specified in the pre-analysis. A more direct comparison between all altruist types and policy consequences on different levels is impractical due to the small sample size of the universal altruist group and the noisiness of the preference elicitation procedure. 32. One may argue that the explanatory power of ethnic and demographic factors is artificially minimized here due to fewer available attributes (compared to material consequences). It is important to note, however, that the effects of immigrants’ number and their sending region were completely negligible in the US MTurk sample (see Figure B10). 33. In a follow-up open-ended question, a few respondents did state that they want to see immigration reduced regardless of the consequences (e.g., “I favor a reduction in immigration, even if it means an impact to household wealth”; “[I] mmigration should be decreased by all means necessary”). In line with my quantitative analysis, however, most respondents emphasized a combination of their personal and national concerns. 34. In a number of related studies, I also demonstrate that the role of parochial altruism generalizes to public preferences on climate change mitigation and electoral behavior across countries (Kustov, 2019a).

References

Baron, J. (2012). Parochialism as a result of cognitive biases. In A. K. Woods, R. Goodman & D. Jinks (Eds.), Understanding social action, promoting human rights. Oxford University Press. Bayram, A. B. (2015). What drives modern Diogenes? Individual values and cosmopolitan allegiance. European Journal of International Relations, 21(2), 451–479. Bechtel, M. M., Genovese, F., & Scheve, K. F. (2019). Interests, norms, and attitudes toward global public goods: The case of climate cooperation. British Journal of Political Science, 49(4), 1–23. Bekkers, R., & Wiepking, P. (2010). A literature review of empirical studies of Philanthropy: Eight mechanisms that drive charitable giving. Nonprofit and Voluntary Sector Quarterly, 40(5), 924–973. Bernhard, H., Fischbacher, U., & Fehr, E. (2006). Parochial altruism in humans. Nature, 442(7105), 912–915. Billig, M. (1995). Banal Nationalism. Sage. Blau, F. D., & Mackie, C. (2016). The economic and fiscal consequences of immigration. The National Academies Press. Böhm, R., Bornstein, G., & Koppel, H. (2014). Between-group conflict and other-regarding preferences in nested social dilemmas. Jena Economic Research Papers, 11, 12–35. Brewer, M. B. (1999). The psychology of prejudice: Ingroup love and outgroup hate? Journal of Social Issues, 55(3), 429–444. Brubaker, R. (2015). Grounds for Difference. Harvard University Press. Buchan, N. R., Brewer, M. B., Grimalda, G., Wilson, R. K., Fatas, E., & Foddy, M. (2011). Global social identity and global cooperation. Psychological Science, 22(6), 821–828. Buntaine, M. T., & Prather, L. (2018). Preferences for domestic action over international transfers in global climate policy. Journal of Experimental Political Science, 5, 1–15. Burns, P., & Gimpel, J. G. (2000). Economic insecurity, prejudicial stereotypes, and public opinion on immigration policy. Political Science Quarterly, 115, 201–225. Chandra, K. (2006). What is ethnic identity and does it matter? Annual Review of Political Science, 9(1), 397–424. Clemens, M. (2011). Economics and emigration: Trillion-dollar bills on the sidewalk? Journal of Economic Perspectives, 25, 83–106. Clemens, M. (2015). Global skill partnerships: A proposal for technical training in a mobile world. IZA Journal of Labor Policy, 4(2), 1–18. Dawes, C. T., Loewen, P. J., & Fowler, J. H. (2011). Social Preferences and Political Participation. The Journal of Politics, 73(3), 845–856. Dickert, S., Västfjäll, D., Kleber, J., & Slovic, P. (2015). Scope insensitivity: The limits of intuitive valuation of human lives in public policy. Journal of Applied Research in Memory and Cognition, 4(3), 248–255. Dinesen, P. T., Klemmensen, R., & Nørgaard, A. S. (2016). Attitudes toward immigration: The Role of personal predispositions. Political Psychology, 37(1), 55–72.

Dreu, C. K. W. D., Balliet, D., & Halevy, N. (2014). Parochial cooperation in humans: Forms and functions of self-sacrifice in intergroup conflict. In A. J. Elliot (Ed.), Advances in motivation science (pp. 1–47). Elsevier. Druckman, J. N. (2014). Pathologies of studying public opinion, political communication, and democratic responsiveness. Political Communication, 31(3), 467–492. Druckman, J. N., & Lupia, A. (2016). Preference change in competitive political environments. Annual Review of Political Science, 19, 13–31. Falk, A., Becker, A., Dohmen, T., Huffman, D., & Sunde, U. (2016). The preference survey module: A validated instrument for measuring risk, time, and social preferences (IZA Discussion Paper No. 9674). Fehr, E., & Schmidt, K. M. (2006). Chapter 8 The economics of fairness, reciprocity and altruism - Experimental evidence and new theories. In S. C. Kolm & J. M. Ythier (Eds.), Handbook on the economics of giving, reciprocity and altruism (pp. 615–691). Elsevier. Gilens, M., & Thal, A. (2018). Doing well and doing good? how concern for others shapes policy preferences and partisanship among affluent Americans. Public Opinion Quarterly, 82(2), 209–230. Goode, J. P., & Stroup, D. R. (2015). Everyday nationalism: Constructivism for the masses. Social Science Quarterly, 96(3), 717–739. Goodwin, M., & Milazzo, C. (2017). Taking back control? Investigating the role of immigration in the 2016 vote for Brexit. British Journal of Politics and International Relations, 19(3), 450–464. Haidt, J. (2016). When and why nationalism beats globalism. Policy: A Journal of Public Policy and Ideas, 32(3), 46. Hainmueller, J., Hangartner, D., & Yamamoto, T. (2015). Validating vignette and conjoint survey experiments against real-world behavior. Proceedings of the National Academy of Sciences of the United States of America, 112(8), 2395– 2400. Hainmueller, J., & Hopkins, D. J. (2014). Public attitudes toward immigration. Annual Review of Political Science, 17, 225–249. Hainmueller, J., & Hopkins, D. J. (2015). The hidden American immigration consensus: A conjoint analysis of attitudes toward immigrants. American Journal of Political Science, 59(3), 529–548. Hainmueller, J., Hopkins, D. J., & Yamamoto, T. (2014). Causal inference in conjoint analysis: Understanding multidimensional choices via stated preference experiments. Political Analysis, 22(2014), 1–30. Hansen, O.-P. M., & Legge, S. (2016). Drawbridges down: Altruism and immigration preferences (CES Working Paper No. 6204). Hassin, R. R., Ferguson, M. J., Shidlovski, D., & Gross, T. (2007). Subliminal exposure to national flags affects political thought and behavior. Proceedings of the National Academy of Sciences of the United States of America, 104, 19757–19761. Herrmann, R. K. (2017). How Attachments to the nation shape beliefs about the world: A theory of motivated reasoning. International Organization, 71(S1), S61–S84.

Mercer, J. J. (2005). Prospect theory and political science. Annual Review of Political Science, 8(1), 1. Milanovic, B. (2015). Global inequality of opportunity: How much of our income is determined by where we live? Review of Economics and Statistics, 97(2), 452– 460. Miller, D., & Ali, S. (2014). Testing the national identity argument. European Political Science Review, 6(2), 237–259. Muste, C. P. (2013). The dynamics of immigration opinion in the United States, 1992- 2012. Public Opinion Quarterly, 77(1), 398–416. Mutz, D. C., & Kim, E. (2017). The impact of ingroup favoritism on trade preferences. International Organization, 71(4), 827–850. Newman, B. J., Hartman, T. K., Lown, P. L., & Feldman, S. (2013). Easing the heavy hand: Humanitarian concern, empathy, and opinion on immigration. British Journal of Political Science, 45(3), 1–25. Norris, P., & Inglehart, R. (2019). Cultural backlash: Trump, Brexit, and authoritarian populism. Cambridge University Press. Oliner, S. P., & Oliner, P. M. (1988). The altruistic personality: Rescuers of Jews in Nazi Europe. Free Press. Penner, L. A., Dovidio, J. F., Piliavin, J. A., & Schroeder, D. A. (2005). Prosocial behaviour: Multilevel perspectives. Annual Review of Psychology, 56(1), 365– 392. Pritchett, L. (2006). Let their people come: Breaking the gridlock on global labor mobility. Center for Global Development. Robalo, P., Schram, A., & Sonnemans, J. (2017). Other-regarding preferences, ingroup bias and political participation: An experiment. Journal of Economic Psychology, 62, 130–154. Rotemberg, J. J. (2014). Models of caring, or acting as if one cared, about the welfare of others. Annual Review of Economics, 6, 129–154. Ruhs, M. (2013). The price of rights: Regulating international labor migration. Princeton University Press. Rusch, H. (2016). Parochial altruism: Pitfalls and prospects. Frontiers in Psychology, 7(1004), 1–3. Schildkraut, D. J. (2014). Boundaries of American identity: Evolving understandings of “Us”. Annual Review of Political Science, 17, 441–460. Schnellenbach, J., & Schubert, C. (2015). Behavioral political economy: A survey. European Journal of Political Economy, 40, 395–417. Schwartz-Shea, P., & Simmons, R. T. (1991). Egoism, parochialism, and universalism: Experimental evidence from the layered prisoners’ dilemma. Rationality and Society, 3(1), 106–132. Sears, D. O., & Funk, C. L. (1990). The limited effect of economic self-interest on the political attitudes of the mass public. Journal of Behavioral Economics, 19(3), 247–271. Singer, P. (2015). The most good you can do: How effective altruism is changing ideas about living ethically. Yale University Press.

Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In W. G. Austin & S. Worchel (Eds.), Psychology of intergroup relations (pp. 7–24). Nelson-Hall. Tankard, M. E., & Paluck, E. L. (2016). Norm perception as a vehicle for social change. Social Issues and Policy Review, 10(1), 181–211. Uhlmann, E. L., Pizarro, D. A., Tannenbaum, D., & Ditto, P. H. (2009). The motivated use of moral principles. Judgment and Decision Making, 4(6), 476–491. van Lange, P., Balliet, D. P., Parks, C. D., & Vugt, M. (2014). Social Dilemmas: Understanding human cooperation. Oxford University Press. Weeden, J., & Kurzban, R. (2017). Self-interest is often a major determinant of issue attitudes. Political Psychology, 38, 67–90. Wong, C. (2010). Boundaries of obligation in American politics: Geographic, national, and racial communities. Cambridge University Press. Yamagishi, T., & Mifune, N. (2008). Does shared group membership promote altruism? Fear, greed, and reputation. Rationality and Society, 20(1), 5–30.

Author Biography

Alexander Kustov is a Postdoctoral Associate in the Jackson Institute for Global Affairs at Yale University. His research focuses on public and policy responses to diversity and migration in high-income democracies. Prior to his appointment at Yale, Alexander received his joint PhD in Politics and Social Policy from Princeton University

Borders of Compassion: Immigration Preferences and Parochial Altruism

Resources

Abstract

Keywords

Introduction

Theory and Hypotheses

A Decision Model of Parochial Altruism in Political Behavior

Implications of Parochial Altruism for Immigration Attitudes

Related and Complementary Explanations

Data and Research Design

Study 1: Why Do Altruistic Voters Resist Immigration?

Observational Design: Elicitation of Parochial Altruism

Analysis and Results

Discussion

Study 2: When Would Most Voters Support Immigration?

Experimental Design: Conjoint Analysis of Policy Choice

Analysis and Results

Discussion

Conclusion

Authors’ Note

Acknowledgments

Declaration of Conflicting Interests

Funding

ORCID iD

Supplemental Material

Notes

References

Author Biography